Purpose: To assess the impact of Strut-Adjusted Volume Implant (SAVI) catheter digitization variability on dosimetric parameters of HDR brachytherapy breast treatment plans.
Methods: Four clinically approved SAVI cases were chosen for this digitization variability analysis study. All patients were implanted with 6-1 SAVI devices. Six experienced physicists independently digitized SAVI catheters. Digitization time was clinically realistic time frame at 25-40 minutes per case. Plans utilizing significant peripheral loading were used for this study where SAVI catheters were in close proximity to chestwall and/or skin. After digitization was complete for each case by each physicist, the original clinical dwell times were copied over to enable the comparison. This exercise ensured that only variability between plans is the digitization of SAVI catheters by different users. The original plan that went through 2 physicist’s checks and physician’s review was considered the ‘ground truth’ plan to which all the other plans were compared. Plans were evaluated on planning parameters for lumpectomy cavity’s PTV_EVAL D90, V150, V200 and for the OARs (chestwall and skin), on D0.03cc, D0.1cc, D1cc, D2cc.
Results: The average and maximum dwell positional digitization uncertainties were 0.36 and 0.75 mm, respectively. Average PTV_Eval D90 was 96.7+/-4.2 %, V150 was 22.4+/-4.3 cc, V200 was 11.5+/-2.0 cc; All OAR constraints were met on all plans – Chestwall and Skin D0.03cc was 103.4+/-9.2 % and 93.6+/-6.0 %, respectively. Two plans lacked coverage (D90) by ~ 3%, a closer look revealed that reoptimization could achieve similar coverage without undermining OAR constraints. Aggregate analysis across all plans shows a clinically non-significant spread around the mean for all parameters considered.
Conclusion: Robustness of SAVI treatment plans to minor variation in catheter digitization was proved through our multi-user study. Our study showed that SAVI planning constraints are stable within reasonable variation of digitization differences.